Slashdot Mirror
Waymo's Driverless Cars Have Logged 10 Million Miles On Public Roads (qz.com)
An anonymous reader quotes a report from Quartz: Alphabet's driverless-car company Waymo announced a new milestone today (Oct. 10): its vehicles have driven a collective 10 million miles on U.S. roads. With cars in six states, Waymo has really been racking up the miles since April 2017, when it launched a program giving rides to passengers around the Phoenix, Arizona area. At that point, Waymo cars had driven not quite 3 million miles since the company's earliest days as a research project within Google in 2009. But in the last 18 months, the company more than tripled its road mileage.
Competing with other companies with autonomous-vehicle programs like Uber, Tesla, Apple, and GM's Cruise, Waymo is leading the pack in terms of road miles driven. [...] The company's next 10 million miles, CEO John Krafcik said in today's announcement, will focus on "striking the balance" between its safety-first algorithms and driving assertively in everyday maneuvers, like merging, and navigating bad weather. But it's worth keeping things in perspective: U.S. drivers rack up some 3 trillion miles each year, so Waymo still has some ground to cover.
Competing with other companies with autonomous-vehicle programs like Uber, Tesla, Apple, and GM's Cruise, Waymo is leading the pack in terms of road miles driven. [...] The company's next 10 million miles, CEO John Krafcik said in today's announcement, will focus on "striking the balance" between its safety-first algorithms and driving assertively in everyday maneuvers, like merging, and navigating bad weather. But it's worth keeping things in perspective: U.S. drivers rack up some 3 trillion miles each year, so Waymo still has some ground to cover.
Even if they put 3 trillion miles on their system, if they confine it to just a few geographical areas, I don't trust it very much. I'd like to see them driving in NYC, Boston, Chicago, New Jersey (even humans can't figure this one out), etc. Places where public investment in the roadways has either been compromised (i.e. stolen by politician for other bullshit), minimal, or there simply wasn't enough space to put proper roads in, so they did something else instead...
Um, huh? Tesla's Autopilot had driven 1,2 billion miles as of July. Two orders of magnitude more than Waymo.
10 million miles is really nothing. In the US, there's only one fatal accident per 86 million miles on average.
"Close the door! What, were you born in a barn?" -- Police chief, "Jesus Christ Supercop"
What would be achieved by Waymo's cars travelling the same number of miles as people-driven cars do ?
What did the writer mean ?
Cause it's another Alphabet product and it's barely reliable as an OS.
Every time when I hear about these crazy miles I've got just one question to ask: has Google finally solved image recognition and I'm not talking about simple cases - I'm talking about deliberate fakes, bad weather conditions, etc. 1, 2, 3.
These issues can easily make your car software make life threatening decisions.
9 steps to greasing your anus for Yoda Doll Insertion!
v 4.98.1
$YodaBSD: src/release/doc/en_US.ISO8859-1/yodanotes/9stepprocess.sgml,v 4.98.1 2017/11/25 13:54:20 tsarkon Exp $
Google's technology works of of their advertising tech. They just know where everyone on Earth is and the car just steers clear. It's like a super duper Pokemon Go.
Any snow in Phoenix lately? How about heavy rain?
But it's worth keeping things in perspective: U.S. drivers rack up some 3 trillion miles each year, so Waymo still has some ground to cover.
Umm, WTF does this have to do with "keeping perspective"? It isn't a competition between Waymo and the rest of us human drivers to see who can rack up the most miles driven.
I've got just one question to ask: has Google finally solved image recognition and I'm not talking about simple cases - I'm talking about deliberate fakes, bad weather conditions, etc.
If they had completely solved such problems one would assume they would be bringing the technology to market for sale. So the answer is obviously that they have solved some problems but not all the problems. Human drivers have problems with bad weather and deliberate fakes too. Although be honest, when was the last time you saw an actual deliberate fake sign? Computers actually could be less susceptible to these since they can reference map databases about what speed limits etc should be for a given location whereas a human driver cannot do that sort of cross referencing while driving.
These issues can easily make your car software make life threatening decisions.
Human drivers make life threatening decisions daily. I'm not really sure what your point is. It makes no difference to a dead person whether the driver that caused the accident was human or computer and to date the human drivers have a FAR larger body count. I think the score is several million dead from human drivers to approximately zero for computer drivers.
Even if they put 3 trillion miles on their system, if they confine it to just a few geographical areas, I don't trust it very much.
Why? Human drivers are demonstrably dangerous and the body count to date heavily favors the computers as the likely safer option. While I'm not suggesting autonomous driving vehicles are ready for prime time yet or that it's a slam dunk that they are safer, I think people like yourself are not really doing a very good job of evaluating the actual risk data. Honestly I don't really trust YOU as a driver either. Nothing personal - you shouldn't trust me either or any other human driver. But the point is that what you should trust is the data and the data so far seems promising with regards to the safety of these things.
But no one is going to see a 65 MPH speed limit sign on a residential road and believe it
Neither will a properly programmed computer. It will know it isn't on a road where that sort of speed is appropriate. In fact the computer probably will be cross referencing in real time what the sign says with one or more GPS maps that tells it what the expected speed limit on a road should be. Human's can't do that sort of real time cross referencing - we just have to use our experience and heuristic problem solving.
Additionally, what they have been doing is a clinical trial on non-volunteers-- everyone who intersects their roadways. That's really really bad.
Disagree. One only has to look at the accident record of human drivers versus autonomous vehicles to see who is currently leading the standings to be the most dangerous. (spoiler: humans have the bigger body count by a wide margin) Quite honestly I trust Waymo more than I trust you (or any other human - it's nothing personal) to operate a vehicle safely based on the available data. My chances of getting killed by a human driver are FAR higher. Anyway it doesn't matter to a dead person whether the driver is human or computer. I didn't consent to you driving your car either for whatever purpose you have either so it's not clear to me how that is in any way logically any different.
In any case I think your analogy to informed consent is incorrect. Informed consent is to ensure you are aware of the dangers you might be facing when you might not be otherwise clear. It also is to prevent people from being coerced into a potentially dangerous therapy involuntarily. That does not apply here. Your decision to go on or near a road with other vehicles is voluntary and I'm quite certain you are well aware of the likely dangers. A computer driving a car instead of a human does not meaningfully alter the nature of the dangers you will face nor will it meaningfully change how you react to them.
I think I read somewhere that the number of miles driven in one single morning commute is something like 130 million miles. So, while 10 million miles is certainly impressive we aren't yet equivalent to 0.4% of a year's commuter driving (2 commutes per day, 5 days a week, 52 weeks per year). And, as previously pointed out, the Waymo miles are not driven in every set of road conditions in the US and are very controlled.
The real question we and they need to discuss is what is the amount of testing and user what conditions so that we as consumers can believe that this technology is well tested? Rain, snow, highways with bridges that ice before the roads do, heavy winds, dense fog, all need to be in the test conditions. Then add cities like NYC, where pedestrians don't yield the right of way and the many other edge conditions, like aggressive drivers doing stupid things.
We have a LONG way to go.
There should be one final test to see if your driverless car makes the cut, or joins the list of wannabes, because only humans can figure this s**t out:
There is a six way intersection, not four, in the vicinity of the Beverly Center Shopping mall (8500 Beverly Blvd, Los Angeles, CA 90048) -- the Northeast corner, if I remember correctly. Everything is fine, and people have been able to shop for designer items on any given day. So go ahead, send your driverless car to the mall.
So, you're receiving its signal, letting you know the car is approaching its destination, and is almost at Cedars-Sinai Medical Center, like 95% done.
Then the lights go out, including the traffic lights. All the drivers halt to figure out WTF? Half the drivers recall the four-way stop at intersections in such situations. But the other half has not. There are tourists too behind the wheel. Your driverless car might sit still indefinitely, waiting for the cows to come home.
First minutes of a real war, and our GPS systems will be down.
https://www.youtube.com/c/BrendaEM
This means that, pretty much, Waymo hasn't even covered a thousandth of a percent of the things it needs to cope with when driving.
Not even remotely true. Clearly they haven't come close to every corner case they could run into (they are still testing after all) but if their tech could handle as little as you claim then they couldn't be operated at all on roads of any description. Furthermore most humans are given a license to drive unsupervised well before they have seen all the problems they can run into. We don't even require that they be full grown adults and the "test" is frankly something of a joke.
And also... that if there is a single accident, that it would scale up 300,000-fold in terms of their overall average accident rate if we all jumped on board them. And Waymo's had quite a few.
That's not how statistical analysis works if you are doing it right. When you have a sample size that small you have to be very careful drawing conclusions about how representative it is of the actual risk. It might easily just be a fluke or some sort of random noise in the data. There would have to be a significant increase in the number of accidents to provide a basis for fair comparison. There have been precisely ZERO fatalities involving Waymo vehicles to date versus approximately 40,000 deaths from human driven vehicles in the US alone in 2017. Granted the Waymo vehicles are operating under more restrictions but I'd call that a promising start.
Uh, Tesla's "autopilot" is a driver assist, not a self-driving vehicle. And it racks up the miles on expressways-- that's the easy kind of driving.
It's also used on surface streets. It can follow cars, stay in complex lanes, and react to changes around it.
I would argue that Tesla has on balance as much important experience as Waymo does, because Tesla has a lot more info on the basics of driving determined in a general purpose way, with no prior knowledge of the road you are on. Waymo's approach is more advanced but can only be used in limited areas where Tesla's approach can in theory be used on any road anywhere. That's a lot more appealing from the standpoint of a consumer buying a car, Waymo's approach is much better in trying to build cars to replace things like taxis in a specific service area... but better hope you don't want to go even ten feet outside that area.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
One could drive 10 trillion miles on a straight country road and would still be ill-equipped to drive in anh major city, around an accident or debris in the road.
Autonomous cars will never happen.
Yaay, Waymo drives 0.0003 percent of the miles that humans do in a year.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
A small risk of accidents and fatalities is an acceptable price for the freedom of driving, as it was for the freedom of riding for those before us. Reducing the risk by taking away the freedom is just not acceptable for many of us.
First off your argument that autonomous cars somehow reduce your freedom is nonsense. The freedom that cars provide is freedom of mobility which is in no way being threatened. If you enjoy driving that's fine but your freedom to drive isn't being impinged by autonomous vehicles also being on the road. Even if computers replaced all human drivers your freedom of mobility isn't being affected at all. The ONLY point is that YOU as a human driver are very likely a bigger threat to me (and vice versa) than these computer driven vehicles so arguments that they are risky is not supported by the data. At no point did I claim that we could/will/should remove human drivers from the roads.
Second, whether it is an acceptable price to pay is a decision we make as a society, not something you get to decide for the rest of us. 40,000 people died last year in the US alone from human driven vehicles and I'm pretty sure some of them and their loved ones might prefer a different outcome if we had the technology. If we replace the human drivers with computers and drop that number to a smaller number (maybe even zero) then you are going to have a VERY hard time arguing that huge body count is a worthwhile price to pay. The purpose of a car is to get from place to place - who is driving it is irrelevant. Just because we've had a body count in the past doesn't mean we need to continue to have one just so we can continue to prove that we humans are shitty drivers. We allow the current situation only because we don't really have a better option currently.
You can live in a padded room if you like, but don't impose it on others.
You misunderstand. I'm not arguing for the padded room. I'm arguing that the perception of autonomous vehicles as risky is not supported by the data so you can probably go out there without worrying about getting mowed down by one. In fact to date the data seems to indicate that they are at least as safe and probably safer than human driven vehicles so if you are being objective about it you should probably worry more about he human drivers than the computer ones.
the US auto fatality rate is 1.5 per 100 million miles. Come back and let us know how things are when you get to a billion miles.
I have yet to be mustered as a Green Beret. Numbers are meaningless if the context within which they are derived is irrelevant. Google is really good at that.
None of this matters. So-called 'self driving cars' still have no capacity to actually think, and so-called 'deep learning algorithms' and so on are not a substitute for actual cognition. I still maintain that these are being rushed to market as fast as they possibly can with the only real goal in mind being to start getting ROI as fast as possible, and their legal departments have assured them that the financial risk of settling lawsuits out of court is acceptable compared to not pushing these out the door as soon as they possibly can. Sadly there will be accidents and deaths before they're banned from public roads, and even then they'll try to lobby and bribe legislators into letting them continue anyway. In the end these will be a massive failure because the 'technology' is not adequate and never will be, no matter how much 'training' you download into it.
I've considered pasting an image of a stop sign on the back of my car, just to see what happens.
It could be a public interest advertising sign, such as:
"[[Stop sign]] think about what self driving cars will do when they see this"
(1) Do any amateur lawyers think I'm taking a legal risk?
(2) Any suggestions on what other signs would be cool to put in front of a Waymo car?
driven during inclement weather?
on snow-covered roads?
in construction zones?
in parking lots?
Waymo's cars can all share the experience they learned from those 10 million miles. Human methods of sharing memories are far, far weaker. So you can roughly divide the human miles by the number of drivers. Very few humans have driven 10M miles in their lifetime.
Do the math and you'd need a whole lot of cars running around all the time at highway speeds. So how many cars are really operating? Something doesn't sound quite right here.
In your own example, half of the humans can't figure it out and you imply that the tourists likewise would have a problem with it. Granted, after a few minutes, everyone figures it out and starts taking turns. I don't think any autonomous vehicle currently being used or on the drawing boards could handle that situation, but that's level 3 and 4 autonomy we're talking about. Level 5 is some years away yet, yet by definition, should be able to handle that situation at least as well, as a human driver.
I need a wheelchair van for my son. Help me get the word out. https://www.gofundme.com/wheelchair-van-for-jj
A robot car could drive around in a 10 mile circle in the desert by itself for thousands of hours... does that make the setup safe for the humans that are not in the desert?